From User Needs Model to AI in Multimodal Content Strategies
Speaker
Executive summary
User needs describe what a piece of content is intended to do for its audience: update, explain, educate, provide perspective, inspire, or serve another specific purpose. When publishers capture that intention alongside topic and precise format, they can see where they are overproducing familiar content and under-serving audience needs.
Dmitry argues that genuine multimodality starts at commissioning. "Video," "audio," and "text" are delivery types, not sufficient formats. An explainer, interview, or analysis can be delivered in any of them. The publisher therefore needs a shared taxonomy across editorial, product, data, and analytics. AI adds another layer: explicit sourcing, attribution, claims, originality, and atomic content structure help machines retrieve and transform information without flattening its significance.
Key takeaways
- User needs turn storytelling intentions into something publishers can define, measure, and improve.
- Topic alone is not enough. The minimum useful structure is topic, user need, and precise format.
- Editorial, product, marketing, and analytics need to describe the same content in the same language.
- Video, audio, and text are delivery types. The format is the editorial structure, such as an explainer, interview, or analysis.
- Multimodality should be designed at commissioning rather than added after a finished story.
- Atomic content blocks allow the same underlying material to be assembled for different audiences, platforms, devices, and moments.
- AI-agent readiness requires clear sourcing, structured claims, attribution, and signals about what is important.
What publishers can do next
- Ask audiences what they are trying to accomplish instead of designing a user-needs model entirely inside the organization.
- Define a minimum controlled vocabulary for topic, user need, and precise content format.
- Align commissioning, CMS fields, product analytics, and performance reporting to that vocabulary.
- Tag a limited collection and test whether the structure improves recommendations, bundles, reuse, or audience insight.
- Add source, claim, attribution, and content-block structure where AI systems need to retrieve or transform material.
- Review whether current dashboards explain only what happened or actively improve the next commissioning decision.
Examples and evidence
- Dmitry presented a user-needs example in which 80% of output generated only 8% of impact.
- He described topic, user need, and precise format as the minimum information required for meaningful multimodal strategy.
- He cited Culture Trip automatically combining separately tagged material into a new article about a "romantic afternoon in south Madrid."
- For fiction, he suggested audience-derived groupings similar to Netflix and Spotify rather than relying only on traditional genre categories.
Important nuance
The User Needs Model originated in news and recurring media publishing. Dmitry's argument is that the underlying logic can apply more broadly, but book publishers should adapt the taxonomy to their readers and products rather than copy a newsroom model unchanged.